Marine and River Dune Dynamics MARID IV 15 & 16 April 2013 - Bruges, Belgium

Gaps in understanding of sedimentary bedforms

in the ancient, the present, the extraterrestrial and the kitchen
M.G. Kleinhans (1)
1. Faculty of Geosciences, Universiteit Utrecht, Utrecht, the Netherlands, m.g.kleinhans@uu.nl
http://www.geo.uu.nl/fg/mkleinhans

Abstract
Much is known about bedforms, but a number of critical gaps in understanding remain. These gaps become clear
when present understanding is extrapolated to extreme materials, conditions or environments. Advanced physicsbased modelling demonstrates that dunes and upper-stage plane bed are predictable from the laws of physics, but
may be sensitive to constitutive relations for flow and particle friction and sediment transport. Furthermore the
question whether current ripples and dunes are distinct forms or the same forms with different sizes remains
unanswered. Empirical work and experiments shows that a zoo of phenomena and bedforms emerge in cohesive
sediments or sand-gravel mixtures, in hyperconcentrated flows and turbidity currents, showing that mixtures of
particle sizes with different weight or other properties require basic exploration. Finally, the angle of repose is the
most basic characteristic of sediment relevant for inter-particle friction and the steep lee side of bedforms, but
prediction from physics remains poorly understood as demonstrated by measurements in lower gravity. The above
phenomena arise because of three general mechanisms that can generally be called friction and are incompletely
understood: 1) the laminar to turbulent flow transition, 2) friction between particles and 3) cohesive forces.

1) Deduction: laws of physics and numerical

physically-based models are applied to
initial+boundary conditions to predict bedforms.
2) Induction: initial+boundary conditions are
measured or experimentally controlled to map the
resulting bedforms to create laws.
3) Abduction: the end result is observed and laws
are considered known, but the formative
conditions are inferred.

1. INTRODUCTION
Bedforms and bed states such as Upper Stage
Plane Bed or Upper Flow Regime Plane Bed have
always been ubiquitous in many environments on
Earth but our understanding of their formation,
dynamics and sedimentary products is patchy.
Classically bedforms and plane bed have been
described for pure sand with low concentrations
under unidirectional currents, tides and waves.
Other species of bedforms emerge in cohesive
sediments or poorly sorted sediment and in starkly
different environments such as hyperconcentrated
surface flows and turbidity currents. Finally,
bedforms were discovered on Mars where a most
fundamental condition differs from that on Earth:
gravitational acceleration is only 3.74 m/s2, yet we
have no clue how this affects bedforms.
The objective of this paper is to uncover critical
gaps in our mechanistic understanding of
bedforms. The method is to explore the extreme
limits of applicability of current concepts for
material properties, conditions and environments.
Building on the value of empiricism and physics,
three complementary styles of logical reasoning:
deduction, induction and abduction are used:

2.
2.1.

PHYSICS-BASED MODELLING
Modelling ripples, dunes and plane bed

A decade ago the physics of full development of

bedforms (as opposed to incipient bedforms
predicted in linear stability analyses) were deemed
so complicated that some doubted a model would
ever be possible. Recently Nabi (2011) modelled
three-dimensional flow and particle motion in
great physical detail to show that ripples, dunes
and upper stage plane bed emerge autonomously.
Furthermore it produces hysteretic dune
development in discharge waves, which can be
used to develop new constitutive relations for dune
dimensions and drag to use in large-scale
morphological engineering models.

Marine and River Dune Dynamics MARID IV 15 & 16 April 2013 - Bruges, Belgium
The physical nature of the model suggests that
these exciting results are universally valid, but the
sediment transport description implemented in the
model is necessarily semi-empirical. Indeed,
alternative constitutive sediment transport relations
did not form dunes. This raises the question what
the effect would be of alternative physical
mechanisms and constitutive relations. Vice versa,
alternative choices can provide clues about the best
forms that constitutive relations for transport could
have, independent of empirical measurement.

1994) show that equilibrium ripples have welldefined dimensions independent of flow velocity,
water depth and particle sizes (except close to the
boundaries of their stability field), namely 0.02 m
high and 0.2 m with a spread of a factor of two.
Current work is aimed at understanding the
relation between scour holes and ripples and the
consequence for the ripple-dune transition.

2.2.

Sorting of sand and gravel in dunes

In seemingly simple unidirectional flow with

sandy gravel several sorting patterns occur, such as
fining upward in dunes and transverse sorting in
river bends. The sorting has large effects on
equilibrium morphology and time scales of
adaptation (Blom et al. 2008), which is highly
relevant for managed rivers such as the Rhine.
Furthermore it interferes with sorting patterns at
the scale of bars and bends. But grain-scale sorting
on reach-scale river morphology pose a rather
large challenge to numerical modelling.
Blom et al. (2008) present a sophisticated model
that conserves mass of all grain size fractions,
models vertical sediment fluxes, and accounts for
stochastic temporal variability of dune trough
depth. This model constitutes an important
improvement over the classic active layer depth
where the layer thickness must be calibrated and
strongly determines time scales of morphological
and or sorting adaptation. The model also points
towards the importance of initial sorting conditions
and history effects in sorting, which adds a
complication to hysteretic bedform dynamics.

Figure 1. Bedform height (left) and length (right)

development in the Nabi (2011) model under constant
conditions starting from flat bed.

A long standing dispute is whether dunes and

ripples are distinct features or simply different in
scale. In bedform stability diagrams ripples occur
in hydraulically smooth bed conditions, i.e. when
laminar sublayer thickness exceeds particle size. In
such conditions local turbulence production over
an irregularity or bedform crest locally enhances
shear stress. This causes formation of a scour hole
which maintains local turbulence strong enough to
suppress the laminar sublayer. If this is true then
dunes replace ripples only due to an increase of
shear stress or grain size and cannot occur in
superposition unless in the transition zone. On the
other hand, these scour holes are also observed
without ripples in near-critical flow, where they
add unrealistic effects to scale models for rivers
and tidal systems. This suggests that scour holes
and rhythmic bedforms are independent forms that
are superimposed in the case of ripples, which
otherwise are the same phenomenon as dunes.
A model such as Nabis (2011) can be used to
remove or add model components representing
physical effects and processes, and can be used to
model scenarios. Modelling started from plane bed
for a flume-sized domain shows that at small
bedforms develop that Nabi calls ripples, which
then merge to grow into dunes. Indeed dune height
growth temporarily stalls during the transition
(Fig. 1). However, the most systematic set of
controlled experiments producing ripples (Baas

Figure 2. Vertical sorting by dunes (Blom et al. 2008).

One important question is what the effect is of

constitutive relations for sediment transport.
Several mechanisms must be included semiempirically. Sediment pickup and deposition must
be corrected for hiding effects occurring with
particles of different sizes. Some processes are too
poorly understood to be included. Due to hiding,
armouring occurs in a mobile layer of a few grains
thickness, whereas in partly mobile sediments the

Marine and River Dune Dynamics MARID IV 15 & 16 April 2013 - Bruges, Belgium
tap water as interstitial fluid. Materials with
angular grains had timeaveraged angles of about
40 and with rounded grains about 25 for all
effective gravitational accelerations, except the
finest glass beads in air, which was explained by
static electricity. For all materials, the static angle
of repose increases about 5 with reduced gravity,
whereas the dynamic angle decreases with about
10. Surprisingly, both depend on gravity and
avalanche size increases with reduced gravity.

smallest particles contributing to morphological

change may be winnowed out of an otherwise
nearly static bed by turbulence penetrating the bed,
or deposited and percolated into the deeper bed.
A second major question is to what extent particlescale phenomena are relevant for reach-scale river
morphology. Yet the thin layers used in the
discretised numerical model require very small
time
steps
unsuitable
for
large-scale
morphodynamic modelling. Yet dunes modify
near-bed flow and thus interfere with transport and
sorting processes on transverse bed slopes on bars
and in bends. These phenomena call for further
study by combination of detailed sorting models,
engineering models and dedicated flume
experiments to unravel the dynamic interaction
between dunes and bars and the transverse slope
effect on sediment transport and sorting.

3.
3.1.

Figure 3. Typical angles of repose under reduced g.

EMPIRICAL EXTREMES
Qualitative explanations are that friction is lower
in more dilated grain flows in reduced gravity so
that the dynamic angle of repose is smaller. The
static angle of repose, on the other hand, is related
to cohesive forces including Vanderwaals and
electrostatic forces, which continue to act in
vanishing gravity and increase the static angle.
We lack the mechanistic understanding to predict
angle of repose and friction angle depending on
material properties (and gravity). Hence this
causes empirical uncertainty in transport
predictors. It also raises questions about what
happens in bedforms with low density sediments
such as shell hash and mud clasts. Furthermore it is
the question which angle is likely to be preserved
at the planetary surface and in the rock record.
Ubiquitous disturbances render the lower dynamic
angle of repose dominant in nature. Lower slip
face angles reduce flow separation and turbulence
generated by flow over bedforms (Nabi 2011).

Angle of repose dependence on g

Noncohesive granular materials are found in many

contexts, from kitchen to industry and, notably, the
lee side of ripples, dunes and other steep bedforms.
A basic property is the angle of repose a: the
maximum slope angle at which the material is at
rest. Above this slope angle, the material starts to
flow. The angle is related to the friction angle of
sediment that figures in semi-empirical relations
for incipient sediment motion and transport.
It is generally believed that this angle is
independent of gravitational acceleration g as on
other terrestrial planets, moons and asteroids in the
Solar system. This belief follows from the
Coulomb law and the first law of Amontons. It
follows that the driving gravitational force along
the slope of a granular flow, Fz = mgeff sina, is
balanced by friction, which depends on the force
normal to the slope, Ff = mgeff cosa. As both scale
with the weight of the flow, the dynamic angle of
repose for a granular flow is independent of
gravity. Although some observations were done
for granular materials under hypergravity, direct
measurements of the angle of repose are rare.
In 33 parabolic flights in a wellcontrolled research
aircraft Kleinhans et al. (2011) recorded
avalanching granular materials in rotating drums at
effective gravitational accelerations of 0.1 and 0.38
times the terrestrial value and at 1.0g as a control
experiment. The granular materials varied in
particle size and rounding and had Utrecht air or

3.2.

Bedforms in sand-mud mixtures

Mud occurs in abundance in fluvial and coastal

systems since the emergence of plant life on Earth
~400 million years ago. Although mud consists of
much finer particles than sand or of low-density
flocs with much lower settling velocity, mud and
sand often mix. Mud adds a plethora of effects to
sediment, including cohesion and reduced pore
flow due to blocking. Furthermore mud modifies
near-bed turbulence. In novel experiments Baas et

Marine and River Dune Dynamics MARID IV 15 & 16 April 2013 - Bruges, Belgium
al. (2011) found a bewildering variety of bedforms
and layer patterns in sand-mud mixtures relevant
for tidal environments and turbidity currents.
Baas et al. found that bedforms generated in
turbulent flow and in turbulence-modulated,
cohesive flows differ greatly in their size, texture,
sedimentary structure and migration rate as a
function of the duration of the formative flow and
the texture of the initial flat bed. In particular, the
mud, silt and sand fractions mix, segregate or layer
rhythmically depending on Reynolds number and
near-bed mud concentration. These findings are
relevant for mud-silt-sand stratification formed in
tidal environments, where critical shear stress for
motion will be affected, and for rapidly
decelerating turbidity currents in deep water.

4.
4.1.

and phases in a turbidity current system and

provide detailed information on the formative
conditions. This work illustrates that the geologic
record may contain a zoo of bedforms that are
hitherto unrecognised and poorly understood.

5.

CONCLUSIONS

The science of bedforms is ultimately founded on

classical physics, but three related groups of
mechanisms are poorly understood and yet cause a
plethora of phenomena highly relevant to the field:
1. laminar-turbulent flow transitions
2. friction between static and mobile
particles, particularly of different sizes
3. cohesion: a bulk term for attractive forces
between particles of clay-size and larger.
The examples further illustrate that complementary
approaches in the disciplines studying bedforms
reveals where significant progress can be made.

BEDFORMS CARVED IN ROCK

Hyperconcentrated supercritical flow

The classical Bouma sequence of layering in rocks

has been used as a conceptual model to infer the
formative conditions of turbidity currents. But new
systematic experimentation in supercritical
hyperconcentrated flow demonstrated that
antidunes, breaking antidunes, chutes-and pools
and, notably, large, upslope migrating cyclic steps
are transitional into each other with increasing
Froude numbers (Cartigny et al. in review). In
particular such large cyclic steps were hitherto
unrecognised bedforms, that now appear to have
recognisable stratification in outcrops of turbidity
currents. Point characterisation of the flow with
migrating bedforms inevitably covers both
bedform troughs and crests, and experimentally it
was found that a high percentile of the Froude
number distribution over time discriminated better
than the median between the bedform patterns.
Temporally stable bedform types (in dynamic
equilibrium) occupy distinct fields in a twodimensional phase diagram with sediment mobility
and particle size. However, the fields are modified
by sediment concentration and fall-out rate.
Cartigny et al. find that various patterns of
stratification resembling hummocky crossstratification can be caused by multiple completely
different bedforms and conditions. Fortunately,
detailed differences exist so that otherwise vague
deposits hitherto interpreted as the result of one
process can now be ascribed to distinct locations

6.

ACKNOWLEDGMENT

I am grateful to Astrid Blom, Mohamed Nabi,

Jacco Baas and George Postma for discussion.
Presented views are entirely my responsibility.

7.

REFERENCES

Baas, J.H. 1994. A flume study on the development and

equilibrium morphology of small-scale bedforms in
very fine sand. Sedimentology, 41, 185209.
Baas, J.H., Best, J.L. & Peakall, J. 2011. Depositional
processes, bedform development and hybrid bed
formation in rapidly decelerated cohesive (mud
sand) sediment flows. Sedimentology 58: 1953-1987
Blom, A., Ribberink, J.S. & Parker, G. 2008. Vertical
sorting and the morphodynamics of bed formdominated rivers: A sorting evolution model. J.
Geophys. Res., 113: F01019
Cartigny, M.J.B., Ventra, D., Postma G. and Van Den
Berg, J.H. submitted. Morphodynamics and
sedimentary structures of bedforms under
supercritical-flow conditions: new insights from
flume experiments. Sedimentology, also open access
PhD thesis Utrecht University 2012.
Kleinhans, M.G., Markies, H. de Vet, S.J. in t Veld,
A.C. & Postema, F.N. 2011. Static and dynamic
angles of repose in loose granular materials under
reduced gravity. J. Geophys. Res. 116: E11004
Nabi, M. 2012. Computational modelling of small-scale
river morphodynamics. Open access PhD thesis
Delft University of Technology, 224 p.